Targeting cIAP2 in a novel senolytic strategy prevents glioblastoma recurrence after radiotherapy

Abstract

Glioblastomas (GBM) are routinely treated with high doses of ionizing radiation (IR), yet these tumors recur quickly, and the recurrent tumors are highly therapy resistant. Here, we report that IR-induced senescence of tumor cells counterintuitively spurs GBM recurrence, driven by the senescence-associated secretory phenotype (SASP). We find that irradiated GBM cell lines and patient derived xenograft (PDX) cultures senesce rapidly in a p21-dependent manner. Senescent glioma cells upregulate SASP genes and secrete a panoply of SASP factors, prominently interleukin IL-6, an activator of the JAK-STAT3 pathway. These SASP factors collectively activate the JAK-STAT3 and NF-κB pathways in non-senescent GBM cells, thereby promoting tumor cell proliferation and SASP spreading. Transcriptomic analyses of irradiated GBM cells and the TCGA database reveal that the cellular inhibitor of apoptosis protein 2 (cIAP2), encoded by the BIRC3 gene, is a potential survival factor for senescent glioma cells. Senescent GBM cells not only upregulate BIRC3 but also induce BIRC3 expression and promote radioresistance in non-senescent tumor cells. We find that second mitochondria-derived activator of caspases (SMAC) mimetics targeting cIAP2 act as novel senolytics that trigger apoptosis of senescent GBM cells with minimal toxicity towards normal brain cells. Finally, using both PDX and immunocompetent mouse models of GBM, we show that the SMAC mimetic birinapant, administered as an adjuvant after radiotherapy, can eliminate senescent GBM cells and prevent the emergence of recurrent tumors. Taken together, our results clearly indicate that significant improvement in GBM patient survival may become possible in the clinic by eliminating senescent cells arising after radiotherapy.

Keywords: Glioblastoma Recurrence; Radiation Therapy; Senescence-associated Secretory Phenotype; Senolytic Therapy; Therapy-induced Senescence.

Figure 1. Irradiated GBM cells undergo senescence, upregulate SASP genes, and secrete SASP factors.

(A) Representative images of SA-β-gal staining of GBM cell lines mock-irradiated (Mock) or irradiated (IR) with 10 Gy of X-rays and then allowed to recover for 10 days (n = 3 with at least 100 nuclei scored for each replicate). Nuclei are stained with DAPI (blue). Plots show mean percentage of SA-β-gal-positive cells +/− SD. A two-tailed Student’s t test was performed; LN229 P = 0.000009, A172 P = 0.000027, U118 P = 0.000141, U87 P = 0.007801. Scale bar, 100 μm. (B) Whole cell extracts from mock-irradiated or irradiated GBM cell lines were western blotted with anti-p21 antibody. Actin serves as loading control. (C) Volcano plot representation of differentially expressed genes from mock-irradiated vs. irradiated LN229 or A172 cells (n = 3), as assessed by RNA sequencing 10 days after treatment with 10 Gy of X-rays. Differential gene expression analysis was performed using edgeR. Gene-specific dispersions were estimated using a tagwise dispersion model and statistical significance was assessed using the likelihood ratio test (LRT). The false discovery rate (FDR) was calculated using the Benjamini–Hochberg procedure. Genes with an FDR < 0.05 with log₂ fold change (Log₂ FC) cutoff of −1 and 1 were considered differentially expressed, as denoted by dashed lines. Core SASP genes that were further validated by qRT-PCR are represented by numbers. (D) Total RNA was isolated from GBM cells 10 days after irradiation with 10 Gy of X-rays or from mock-irradiated cells and expression of SASP genes assessed by qRT-PCR (n = 3 biological replicates comprising three technical replicates each). Plots show mean fold change (Log₂ FC) in gene expression +/− SD of SASP-related genes in irradiated GBM cells relative to mock-irradiated cells. A two-tailed Student’s t test was performed; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (please refer to Appendix Table S4 for the exact p values). (E) Conditioned media were collected from GBM cells 10 days after irradiation with 10 Gy of X-rays (n = 3 biological replicates comprising three technical replicates each), and cytokine levels were measured by multiplex bead-based immunoassay. Plots show mean fold change (Log₂ FC) in cytokine levels +/− SD in media from irradiated cells relative to mock-irradiated cells. A two-tailed Student’s t test was performed; **P < 0.01; ***P < 0.001; ****P < 0.0001 (please refer to Appendix Table S4 for the exact P values). Source data are available online for this figure.

Figure 2. Senescent GBM cells activate the JAK-STAT3 pathway in naive GBM cells.

(A) Naive GBM cells were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) GBM cells for the indicated times, and activation of the JAK-STAT3 pathway assessed by western blotting with anti-phopho-STAT3 (Y705) antibody. Actin serves as loading control. Recipient cells were serum starved (ss) before addition of CM. (B) Phosphorylation of STAT3 after exposure of naive cells to CM-SEN for 30 min was assessed by western blotting after either neutralization of IL-6 in the conditioned media with an anti-IL-6 antibody (N-α-IL-6) or treatment of recipient GBM cells with the JAK inhibitor ruxolitinib, as indicated. (C) Serum starved (ss) GBM cells were pulsed with BrdU after exposure to CM-NS, CM-SEN, or CM-SEN in the presence of ruxolitinib (CM-SEN+Ruxo), and immunofluorescence stained with anti-BrdU antibody (red), as indicated (n = 3 with at least 100 nuclei scored for each replicate). Nuclei are stained with DAPI (blue). Plots show mean percentages of BrdU-positive cells +/− SD. A two-tailed Student’s t test was performed; LN229 P = 0.00183, 0.0028; A172 P = 0.01061, 0.03826; U118 P = 0.00736, 0.02480; U87 P = 0.00002, 0.00002, respectively. Scale bar, 10 μm. Source data are available online for this figure.

Figure 3. Senescent GBM cells upregulate the anti-apoptotic gene BIRC3 and induce BIRC3 in naive cells.

(A) List of BCL-2 and IAP family members showing fold changes (Log₂ FC) in gene expression in irradiated (IR) LN229 or A172 cells relative to mock-irradiated (mock) cells along with mRNA transcripts levels in counts per million (CPM), as assessed by RNA sequencing 10 days after irradiation with 10 Gy of X-rays (n = 3). Genes with Log₂ fold change greater than 3 are highlighted in red. (B) MA plot showing abundance of transcripts represented by Log₂ CPM (y axis) vs. Log₂ fold change in gene expression (x axis) in irradiated cells relative to mock-irradiated cells (n = 3 biological replicates per condition for each cell line). BCL-2 and IAP family gene members are denoted with numbers. Genes with log₂ fold change (Log₂ FC) cutoff of −1 and 1, and Log₂ CPM higher than −1 were considered differentially expressed, as denoted by dashed lines. (C) Kaplan–Meier curve showing correlation of higher BIRC3 expression levels with poor prognosis in GBMLGG patients in TCGA (n = 606) and CGGA (n = 657) cohorts as evidenced by Hazard Ratio (logrank) of 3.441 and 1.811, respectively. P < 0.0001 for both plots. (D) Plot shows mean relative expression of BIRC3 + /− SD in mock-irradiated vs. irradiated (IR) GBM cells 10 days after exposure to 10 Gy of X-rays, as assessed by qRT-PCR (n = 3 biological replicates comprising three technical replicates). A two-tailed Student’s t test was performed; exact P values from left to right: 0.00001, 0.00048, 0.00397, 0.00012. (E) Whole cell extracts from mock-irradiated or irradiated GBM cell lines were western blotted with anti-cIAP2 antibody. Actin serves as loading control. (F) Senescent LN229 cells (10 days after exposure to 10 Gy) were treated with the IKK inhibitor BMS-345541 (BMS) or DMSO as control for 72 h (n = 3 biological replicates comprising three technical replicates), and mean relative expression of BIRC3 + /− SD was assessed by qRT-PCR (a two-tailed Student’s t test was performed; P = 0.00000002, 0.00000003, respectively) or (G) western blotting for cIAP2. (H) Naive LN229 cells were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) cells for the indicated times, and expression of cIAP2 assessed by western blotting. (I) Naive LN229 cells were treated with BMS-345541 or DMSO as control for 2 h before exposure to CM-SEN (n = 3 biological replicates comprising 3 technical replicates) and mean relative expression of BIRC3 + /− SD was assessed by qRT-PCR (a two-tailed Student’s t test was performed; P = 0.00001, 0.00002, respectively) or (J) western blotting for cIAP2. Source data are available online for this figure.

Figure 4. The SMAC mimetic birinapant selectively eliminates senescent GBM cells.

(A) GBM cells were irradiated with 10 Gy of X-rays (IR) and treated with the cIAP2 inhibitor birinapant (or DMSO as control) after 10 days for the indicated times, and cIAP2 levels and cleavage of caspase-8 were assessed by western blotting. Actin serves as loading control. (B) Mock-irradiated or irradiated GBM cells were treated with birinapant or DMSO as control, and the surviving cells were visualized by staining with crystal violet at the indicated times (Scale bar, 500 μm), and (C) viability (normalized to that of DMSO-treated cells, n = 8 or 9 replicates per cell line) was quantified by the MTT assay. The drug was replaced every 72 h. Plots show mean viability +/− SD. Source data are available online for this figure.

Figure 5. Senescence, upregulation of SASP genes, and paracrine effects in GBM PDX cultures.

(A) Representative images of SA-β-gal staining of GBM PDX cultures mock-irradiated (Mock) or irradiated (IR) with 10 Gy of X-rays and then allowed to recover for 10 days (n = 3 with at least 100 nuclei scored for each replicate). Nuclei are stained with DAPI (blue). Plots show mean percentages of SA-β-gal-positive cells +/− SD. A two-tailed Student’s t test was performed; GBM6 P = 0.0064832, GBM12 P = 0.0001251, GBM43 P = 0.0000002, GBM123 P = 0.0001289, GBM148 P = 0.0000007, GBM245 P = 0.0000155). Scale bar, 50 μm. (B) Whole cell extracts from mock-irradiated or irradiated PDX cultures were western blotted with anti-p21 and anti-p16 antibodies, as indicated. Actin serves as loading control. (C) Total RNA was isolated from PDX cultures 10 days after irradiation with 10 Gy of X-rays, and expression of SASP genes assessed by qRT-PCR (n = 3 biological replicates comprising three technical replicates each). Plots show mean fold change (Log₂ FC) in gene expression +/− SD of SASP-related genes in irradiated GBM cells relative to mock-irradiated cells. A two-tailed Student’s t test was performed; ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (Please refer to Appendix Table S4 for the exact P values). (D) Naive PDX cultures were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) PDX cultures for the indicated times, and activation of the JAK-STAT3 pathway assessed by western blotting with anti-phopho-STAT3 (Y705) antibody. Actin serves as loading control. Recipient cells were serum starved (ss) before addition of CM. Source data are available online for this figure.

Figure 6. cIAP2 induction and birinapant sensitivity in GBM PDX cultures.

(A) GBM12 cells were irradiated with 10 Gy of X-rays and treated with the cIAP2 inhibitor birinapant (or DMSO as control) after 10 days for the indicated times, and cIAP2 levels and cleavage of caspase-8 were assessed by western blotting. Actin serves as loading control. (B) Whole cell extracts from mock-irradiated or irradiated GBM PDX cultures treated with DMSO or birinapant for 2 h were western blotted with anti-CIAP2 antibody. (C) Plots show mean relative expression of BIRC3 + /− SD in mock-irradiated (Mock) vs. irradiated (IR) PDX cultures 10 days after exposure to 10 Gy of X-rays, as assessed by qRT-PCR (n = 3 biological replicates comprising three technical replicates). A two-tailed Student’s t test was performed; GBM6 P = 0.00004, GBM12 P = 0.00001, GBM43 P = 0.00015, GBM123 P = 0.00053, GBM148 P = 0.00026, GBM245 P = 0.00016. (D) Naive PDX cultures were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) cultures for the indicated times, and expression of cIAP2 assessed by western blotting. (E) Mock-irradiated or irradiated PDX cultures were treated with birinapant or DMSO as control (n = 3–5 replicates per PDX) and viability was quantified by the MTT assay (normalized to that of DMSO-treated cells). The drug was replaced every 72 h. Plots show mean viability +/− SD. Source data are available online for this figure.

Figure 7. Birinapant delays recurrence in pre-clinical mouse GBM models.

(A) GBM12 cells were injected intracranially in nude mice to generate orthotopic brain tumors (n = 3 mice per treatment group). Mice were mock-irradiated (Mock) or cranially irradiated (IR) with X-rays (10 Gy) and sacrificed after 5 days. Representative images of SA-β-gal staining of tumor cryosections are shown (numbers denote mouse IDs). Scale bar, 20 μm. (B) GBM12 cells expressing firefly luciferase were injected intracranially in nude mice to generate orthotopic brain tumors that were monitored by BLI (n = 9 mice per treatment group). Mice with established tumors were randomized into the following treatment groups: (i) mock treatment (Mock), (ii) birinapant alone (Birinapant), (iii) ionizing radiation alone (IR), and (iv) IR with adjuvant birinapant (IR+Birinapat). BLI images show tumor progression in a representative mouse for each treatment arm. (C) Survival of brain tumor-bearing mice was recorded and represented in a Kaplan–Meier plot. n = 9 mice per group. P = 0.004 (IR vs IR + Birinapant). (D) Mice with GBM12 tumors (n = 3 mice per group) were cranially irradiated with X-rays and treated 5 days later with 10 cycles of birinapant or vehicle, as indicated, and sacrificed 24 h after the last treatment. Representative images of SA-β-gal staining of tumor cryosections are shown. Scale bar, 20 μm. (E) GL261 cells were injected intracranially in C57BL/6J mice to generate orthotopic brain tumors (n = 3 mice per treatment group). Mice were irradiated with X-rays (10 Gy) and sacrificed after 5 days. Representative images of SA-β-gal staining of tumor cryosections are shown. Scale bar, 20 μm. (F) GL261 cells expressing firefly luciferase were injected intracranially in C57BL/6 J mice to generate orthotopic brain tumors that were monitored by BLI (n = 12 mice per treatment group). Mice with established tumors were randomized into the indicated treatment groups. BLI images show tumor progression in a representative mouse for each treatment arm. (G) Survival of brain tumor-bearing mice was recorded and represented in a Kaplan–Meier plot. n = 12 mice per group. P = 0.005 (IR vs IR + Birinapant). (H) Mice with GL261 tumors (n = 3 mice per group) were cranially irradiated with X-rays and treated 5 days later with nine cycles of birinapant or vehicle, as indicated, and sacrificed 24 h after the last treatment. Representative images of SA-β-gal staining of tumor cryosections are shown. Scale bar, 20 μm. Source data are available online for this figure.

Figure EV1. Senescent GBM cells secrete SASP factors that can potentially activate the JAK-STAT3 and NF-kB pathways.

(A) Representative images of GBM cell lines immunofluorescence stained for Lamin B1 (green) and (B) Ki67 (green), 10 days after irradiation (IR) with 10 Gy of X-rays or mock-irradiation (Mock). n = 3 with at least 100 nuclei scored for each replicate. Nuclei are stained with DAPI (blue). Plots show mean percentages +/− SD of Lamin B1- or Ki67-positive cells. A two-tailed Student’s t test was performed; Lamin B1 - LN229 P = 0.00000163109, A172 P = 0.00000004724, U118 P = 0.00000000002, U87 P = 0.00000702223; Ki67 - LN229 P = 0.000030, A172 p = 0.000002, U118 p = 0.000028, U87 P = 0.000005. Scale bar, 50 μm. (C) Heatmap of top 100 differentially expressed genes in LN229 or A172 GBM cells mock-irradiated (Mock) or irradiated (IR) with 10 Gy of X-rays and then allowed to recover for 10 days (n = 3), as assessed by RNA sequencing. (D) Senescence scores of mock-irradiated or irradiated LN229 and A172 cells (n = 3) generated by analysis of RNA-seq datasets using the SenCan Classifier tool. Score ranges from 0 (no senescence) to 1 (senescence). Plot shows mean senescence score +/− SD for both GBM cell lines. A two-tailed Student’s t test was performed; LN229 P = 0.00002368169637, A172 p = 0.00000000000002. (E) Genes involved in SASP, Cytokine-cytokine receptor interaction, JAK-STAT signaling, and NF-kB signaling are significantly enriched in irradiated cells compared to mock-irradiated cells. Normalized Enrichment Scores (NES) are shown in the figure; P = 0. (F) SASP-transcription factor network generated by the list of 24 common cytokines secreted by all four senescent GBM cell lines cross-referenced against the TRRUST database. The visualization was generated using Cytoscape version 3.9.1. Transcription factors are shown in green. The cytokines not shown had no available information in the database. Source data are available online for this figure.

Figure EV2. Senescent GBM cells activate the NF-kB pathway in naive GBM cells.

(A) Naive GBM cells were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) GBM cell lines for the indicated times, and activation of the NF-kB pathway assessed by western blotting with anti-phopho-p65 (S536) antibody. Actin serves as loading control. Recipient cells were serum starved (ss) before addition of CM. (B) Serum starved GBM cells were exposed for 2 h to CM-NS or CM-SEN, and expression of SASP-related genes (relative to expression levels in serum starved cells) was assessed by qRT-PCR (n = 3 biological replicates comprising 3 technical replicates each). Plots show mean fold change in gene expression +/− SD of SASP-related genes in CM-SEN-treated cells relative to CM-NS-treated cells. A two-tailed Student’s t test was performed; ns, not significant; *P < 0.05; **P < 0.01; ****P < 0.0001 (please refer to Appendix Table S4 for the exact P values). Source data are available online for this figure.

Figure EV3. Senescent GBM cells induce BIRC3 in naive cells and promote resistance to ionizing radiation.

(A) Kaplan–Meier curve showing lack of correlation of Bcl2l10 expression levels with prognosis in GBMLGG patients in TCGA (n = 606) and CGGA (n = 657) cohorts as evidenced by Hazard Ratio (logrank) of 0.9692 and 0.8629 and P values of 0.1943 and 0.2640, respectively. (B) Plot shows mean relative expression of BIRC2 + /− SD in mock-irradiated vs. irradiated (IR) GBM cells 10 days after exposure to 10 Gy of X-rays, as assessed by qRT-PCR (n = 3 biological replicates comprising 3 technical replicates each). A two-tailed Student’s t test was performed; exact P values from left to right: 0.0000001, 0.0010112, 0.0044335, 0.9295618. (C) Whole cell extracts from mock-irradiated or irradiated GBM cell lines were western blotted with anti-cIAP1 antibody. Actin serves as loading control. (D) Senescent GBM cells (10 days after exposure to 10 Gy) were treated with the IKK inhibitor BMS-345541 (BMS) or DMSO as control for 72 h (n = 3 biological replicates comprising 3 technical replicates each), and mean relative expression of BIRC3 + /− SD was assessed by qRT-PCR (a two-tailed Student’s t test was performed; A172 p = 0.00000003, 0.00000232; U118 p = 0.00000220, 0.00026887; U87 P = 0.00000818, 0.00118914, respectively) or (E) western blotting for cIAP2. (F) Naive GBM cells were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) cells for the indicated times, and expression of cIAP2 assessed by western blotting. (G) Naive GBM cells were treated with BMS-345541 or DMSO as control for 2 h before exposure to CM-SEN (n = 3 biological replicates comprising 3 technical replicates each), and mean relative expression of BIRC3 + /− SD was assessed by qRT-PCR (a two-tailed Student’s t test was performed; A172 P = 0.000627, 0.000622; U118 P = 0.000001, 0.000001; U87 P = 0.000593, 0.000936, respectively) or (H) western blotting for cIAP2. (I) GBM cells were exposed to conditioned media from senescent or non-senescent cells (n = 3) and radiation sensitivity measured by the colony survival assay. The mean percentage of surviving colonies +/− SD (y axis) is plotted against the corresponding radiation dose (x axis). Source data are available online for this figure.

Figure EV4. The SMAC mimetic LCL161 selectively eliminates senescent GBM cells.

(A) GBM cells were irradiated with 10 Gy of X-rays (IR) and treated with the cIAP2 inhibitor LCL161(or DMSO as control) after 10 days for the indicated times, and cIAP2 levels and cleavage of caspase-8 were assessed by western blotting. Actin serves as loading control. (B) Mock-irradiated or irradiated GBM cells were treated with LCL161or DMSO as control, and the surviving cells were visualized by staining with crystal violet at the indicated times (scale bar, 500 μm), and (C) viability (normalized to that of DMSO-treated cells) was quantified by the MTT assay (n = 6–9 replicates per cell line). The drug was replaced every 72 h. Plots show mean viability +/− SD. Source data are available online for this figure.

Figure EV5. Paracrine effects of senescent GBM PDX cultures and sensitivity to birinapant.

(A) Naive GBM PDX cultures were exposed to conditioned media from senescent (CM-SEN) or non-senescent (CM-NS) PDX cultures for 30 min, and activation of the JAK-STAT3 pathway assessed by western blotting with anti-phopho-STAT3 (Y705) antibody. Actin serves as loading control. Recipient cells were serum starved (ss) before addition of CM and were untreated or treated with the JAK inhibitor ruxolitinib, as indicated. (B) Serum starved (ss) PDX cells were pulsed with BrdU after exposure to CM-NS or CM-SEN in the presence or absence of ruxolitinib (n = 3 with at least 100 nuclei scored for each replicate), and immunofluorescence stained with anti-BrdU antibody. Nuclei are stained with DAPI (blue). Plots show mean percentages of BrdU-positive cells +/− SD. A two-tailed Student’s t test was performed; GBM6 P = 0.00014, 0.00919; GBM12 P = 0.00040, 0.00214; GBM43 P = 0.00004, 0.00069; GBM123 P = 0.00101, 0.00148; GBM148 P = 0.00031, 0.00130; GBM245 P = 0.00191, 0.00072, respectively. (C) Mock-irradiated or irradiated PDX cultures were treated with birinapant or DMSO as control. The drug was replaced every 72 h, and the surviving cells were visualized by staining with crystal violet at the indicated times. Scale bar, 500 μm. Source data are available online for this figure.